The influence of spatial variations in rain intensity for cloudburst modelling : a case study of the Gävle cloudburst

Sammanfattning: With an intensification of heavy rain events in a changing climate and a rapid urbanization the risk for pluvial flooding is increasing in our societies. Pluvial flooding, which is formed when the rainfall rate exceeds the infiltration or drainage rate, can occur rapidly and cause great damages, large economic losses and possibly risk human lives. This kind of flooding is difficult to predict since it is caused by short-term and often local processes, but preventive measures and more robust infrastructure developed over the last decades have decreased the risk of the most severe damages. One way to prevent damage is to map risk areas and take measures by performing a cloudburst modelling, which can be done as a 2D hydraulic modelling. Common practice in cloudburst modelling today is to use a uniform design storm, often the Chicago Design Storm (CDS), with the same hyetograph applied evenly over the whole model area. Even though rain is not spatially uniform this assumption might be valid for more stratiform frontal rain. Intense rain events however have a higher spatial variation in rain intensity, and an assumption like this might significantly affect the results. This study aimed to investigate the effect of the spatial variation in rain intensity on the modelled hydraulic response from an intense rain event. It was performed through a case study of the cloudburst in Gävle, Sweden, in August 2021. A 2D hydraulic model of the city was prepared in the software MIKE 21 Flow Model FM and the cloudburst event was simulated with a spatially varied rainfall input, based on radar data from the event with a 2x2 km resolution, and with spatially uniform rainfall input both with the temporal variation in rain intensity from the event and with a Chicago Design Storm, all with the same total volume. The scenarios were evaluated in terms of proportion of the model area being flooded, the average maximum flooding depth and by mapping the difference in flooding depth over the whole area. The results showed that the spatial variation of rainfall input had a significant effect on the hydraulic response in the city and that assuming a uniform rainfall might lead to an underestimation of the flooding depths in parts of the model area compared to a varied one. The average flooding depth was only a few percent higher for the spatially varied rain compared to the uniform rain with a similar time variation, but in large central areas of the city the model with the uniform rain underestimated the maximum flooding depth by 5-35%. The uniform CDS rain was seen to both over- and underestimate the flooding depth, but in the central and flooded parts of the city underestimation dominated. This points out a risk of using uniform design storms in cloudburst modelling, since a spatially varied rain of the same volume could give more severe effects than the simulated response and that using a uniform design storm potentially introduces an uncertainty in the modelled results that could be important to point out and further quantify.